Background: Genomic intra-tumor heterogeneity (ITH) drives tumor evolution, leading to immune evasion and resistance to therapy. Emerging evidence implicates the transcriptome as a source of important variation that impacts tumor phenotype. Here, we perform a genomic-transcriptomic analysis of intra-tumor transcriptomic diversity upon 941 tumor regions taken from 357 TRACERx non-small cell lung cancers (NSCLC) across primary and metastatic disease subjected to high-depth bulk DNA and RNA sequencing, as well as 91 tumor-adjacent normal tissue samples.

Results: Genomic and transcriptomic diversity are linked across primary and metastatic disease, with expression signatures of proliferation being enriched in the metastasis seeding subclone of the primary tumor relative to non-metastasis seeding subclones. Copy-number independent allele-specific expression (CN-independent ASE), a source of transcriptome-specific diversity, affects 1% (± 0.5%) of genes and is underpinned by aberrant allele-specific methylation (OR=7.58, p≤2.2x10-16), thus providing a window to the NSCLC epigenomic landscape. Driver mutations in chromatin remodellers and histone modifiers, in particular SETD2 and KDM5C, are associated with increased global levels of CN-independent ASE (p=0.0001). In genomically stable tumors, high levels of CN-independent ASE are linked to expression signatures consistent with genomic instability and proliferation (R=0.58, p=0.001), highlighting convergence between the genome and transcriptome in tumor evolution. For the first time, we uncover mutational signatures of RNA editing. Analysis of their activity links the expression of ADAR and APOBEC enzymes to editing processes revealing otherwise hidden APOBEC activity within tumors at sampling (RNA APOBEC activity identified in 188 tumor regions (32%) without evidence of DNA APOBEC activity). RNA editing activity is shared between primary tumors and paired metastasis, but not paired tumor-adjacent normal tissue, suggestive of heritability of this somatic transcriptional process. Finally, we combine multiple measures of genomic and transcriptomic variation in a multi-region approach to define important variation within cancer genes. We illustrate examples that would be missed with a purely genomic focus and demonstrate genomic-transcriptomic parallel evolution, converging on disruption to single cancer genes, such as FAT1 and APC, in different regions of a tumor.

Conclusions: This work highlights the importance of the transcriptome during tumor evolution, as well as the power of integrative multi-omic assessments of ITH, and provides novel insight into the role of transcriptomic variation in lung cancer.

Citation Format: James R. M. Black, Carlos Martinez-Ruiz, Clare Puttick, Jonas Demeulemeester, Elizabeth Larose Cadieux, Kerstin Thol, Thomas P. Jones, Selvaraju Veeriah, Cristina Naceur-Lombardelli, Andrew Rowan, Sophia Ward, Michelle Dietzen, Ariana Huebner, Maise Al Bakir, Miljana Tanic, Thomas B. Watkins, Emilia L. Lim, Ali M. Al Rashed, Daniel E. Cook, Rachel Rosenthal, Gareth Wilson, Alexander M. Frankell, Nnennaya Kanu, Kevin Litchfield, Nicolai J. Birkbak, Allan Hackshaw, Stephan Beck, Peter Van Loo, Mariam Jamal-Hanjani, the lung TRACERx Consortium, Charles Swanton, Nicholas McGranahan. Genomic transcriptomic evolution in TRACERx lung cancer and metastasis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1603.